DSResSol: A Sequence-Based Solubility Predictor Created with Dilated Squeeze Excitation Residual Networks

Protein solubility is an important thermodynamic parameter that is critical for the characterization of a protein’s function, and a key determinant for the production yield of a protein in both the research setting and within industrial (e.g., pharmaceutical) applications. Experimental approaches to predict protein solubility are costly, time-consuming, and frequently offer only low success rates. To reduce cost and expedite the development of therapeutic and industrially relevant proteins, a highly accurate computational tool for predicting protein solubility from protein sequence is sought. While a number of in silico prediction tools exist, they suffer from relatively low prediction accuracy, bias toward the soluble proteins, and limited applicability for various classes of proteins. In this study, we developed a novel deep learning sequence-based solubility predictor, DSResSol, that takes advantage of the integration of squeeze excitation residual networks with dilated convolutional neural networks and outperforms all existing protein solubility prediction models. This model captures the frequently occurring amino acid k-mers and their local and global interactions and highlights the importance of identifying long-range interaction information between amino acid k-mers to achieve improved accuracy, using only protein sequence as input. DSResSol outperforms all available sequence-based solubility predictors by at least 5% in terms of accuracy when evaluated by two different independent test sets. Compared to existing predictors, DSResSol not only reduces prediction bias for insoluble proteins but also predicts soluble proteins within the test sets with an accuracy that is at least 13% higher than existing models. We derive the key amino acids, dipeptides, and tripeptides contributing to protein solubility, identifying glutamic acid and serine as critical amino acids for protein solubility prediction. Overall, DSResSol can be used for the fast, reliable, and inexpensive prediction of a protein’s solubility to guide experimental design.

Exploring Reliability and Factorial Validity of Responsible Talk and its Impact on Cooperative Classrooms

The aim of this study was to validate an instrument which enables the evaluation of talk which maximizes student performance during different segments of interaction-interactivity throughout a complete learning sequence. Based on works developed by the Learning and Research Development Center of the University of Pittsburgh, a scale was developed that gathered the most relevant behaviors of each proposed dimension by researchers from this university center. The scale was used to develop a core subject for a final year Bachelor of Arts degree in Primary Education at a university in Spain and was applied to the 65 students (M = 19, F = 46) taking the subject. The data analysis used an exploratory factor analysis (EFA) that yielded a reliability of α = 0.922. EFA revealed a final interpretable three-factor structure, and the factorial solution comprised 87.86% of total variance. Results show that the talk that students use has three purposes: to constitute an effective group for learning, to build knowledge and to verify its acquisition. The results are discussed in terms of input from the Center for Research in Education and Educational Technologies at the Open University and the Learning Research and Development Center of the University of Pittsburgh.

Learning exceptions to the rule in human and model via hippocampal encoding

Category learning helps us process the influx of information we experience daily. A common category structure is “rule-plus-exceptions,” in which most items follow a general rule, but exceptions violate this rule. People are worse at learning to categorize exceptions than rule-following items, but improved exception categorization has been positively associated with hippocampal function. In light of model-based predictions that the nature of existing memories of related experiences impacts memory formation, here we use behavioural and computational modelling data to explore how learning sequence impacts performance in rule-plus-exception categorization. Our behavioural results indicate that exception categorization accuracy improves when exceptions are introduced later in learning, after exposure to rule-followers. To explore whether hippocampal learning systems also benefit from this manipulation, we simulate our task using a computational model of hippocampus. The model successful replicates our behavioural findings related to exception learning, and representational similarity analysis of the model’s hidden layers suggests that model representations are impacted by trial sequence: delaying the introduction of an exception shifts its representation closer to its own category members. Our results provide novel computational evidence of how hippocampal learning systems can be targeted by learning sequence and bolster extant evidence of hippocampus’s role in category learning.

Literature review as a didactical content analysis: Sustainable magnetic lubricants based on ionic liquids

Magnetic lubricants are usually petroleum-based, but not renewable and cannot be environmentally degraded. It can cause the environmental problems. Magnetic lubricants based on ionic liquids can be environmental friendly. The ionic liquid lubricants are synthesized from vegetable oil fatty acids, which is a locally sustainable and renewable sources. This molecular engineering can be used to integrate the concept of sustainability into teaching and learning. This study aimed to obtain the concept map and teaching learning sequence (TLS) from the scientist’s conception. The method used is a qualitative content analysis (literature analysis type), using an instrument in the form of a content analysis format. The first phase begins with collecting literatures in the form of textbooks, monographs, review results and research articles. The next phase is descriptive analysis, selecting categories, and evaluating the material didactically. This research produces the concept map, TLS and clarified chemical concepts. The scientist’s conception obtained is the application, function, characteristics of media magnetic lubricants, ionic liquids and examples of magnetic lubricants based on ionic liquids. Concept map and TLS can illustrate the relationship between one concept and another. They also show the relationship between science, technology and engineering. The results can be used as the basis for the preparation of teaching materials and didactical designs for teaching and learning.

An Educational Reconstruction of Special Relativity Theory for Secondary Education

Einstein’s derivation of special relativity theory (SRT), based on hypothetical reasoning and thought experiments, is regarded as a prime example of physics theory development. In secondary education, the introduction of SRT could provide a great opportunity for students to engage in physics theorizing, but this opportunity is largely being missed in current teaching practice. One reason could be that secondary students lack some knowledge of electromagnetism that was central to Einstein’s argument. Therefore, we conducted an educational reconstruction to develop a teaching approach that would not rely on advanced understanding of electromagnetism, yet retain the modes of reasoning that were characteristic of Einstein’s approach. In our reconstruction, we identified the light postulate, which is notoriously difficult for students to grasp, as a central concept. We developed a teaching and learning sequence in which students perform relativistic thought experiments and try different interpretations of the light postulate. Through these activities, students experienced how the new concepts meet the requirements for a good theory. Experimental evaluation of the teaching and learning sequence indicates that this can be a fruitful approach to introduce SRT to secondary students.

Connecting Mathematics and Science in Primary School STEM Education: Modeling the Population Growth of Species

In this research, we explored the potential of using a research-based teaching and learning sequence to promote pupils’ engagement in practices that are coherent with those of real world mathematical and scientific activity. This STEM (Science, Technology, Engineering and Mathematis) sequence was designed and implemented by pre-service teachers and science and mathematics education researchers with the aim of modeling the growth of a real population of rabbits. Results show explicit evidence of pupils’ engagement in relevant mathematical and scientific practices, as well as detailed descriptions of mathematical connections that emerged from those practices. We discuss how these practices and connections allowed the progressive construction of models, and the implications that this proposal may have for STEM task design and for the analysis of extra-mathematical connections.

Sequential Recommendation through Graph Neural Networks and Transformer Encoder with Degree Encoding

Predicting users’ next behavior through learning users’ preferences according to the users’ historical behaviors is known as sequential recommendation. In this task, learning sequence representation by modeling the pairwise relationship between items in the sequence to capture their long-range dependencies is crucial. In this paper, we propose a novel deep neural network named graph convolutional network transformer recommender (GCNTRec). GCNTRec is capable of learning effective item representation in a user’s historical behaviors sequence, which involves extracting the correlation between the target node and multi-layer neighbor nodes on the graphs constructed under the heterogeneous information networks in an end-to-end fashion through a graph convolutional network (GCN) with degree encoding, while the capturing long-range dependencies of items in a sequence through the transformer encoder model. Using this multi-dimensional vector representation, items related to the a user historical behavior sequence can be easily predicted. We empirically evaluated GCNTRec on multiple public datasets. The experimental results show that our approach can effectively predict subsequent relevant items and outperforms previous techniques.

DSResSol: A sequence-based solubility predictor created with Dilated Squeeze Excitation Residual Networks

Protein solubility is an important thermodynamic parameter critical for the characterization of a protein's function, and a key determinant for the production yield of a protein in both the research setting and within industrial applications. Thus, a highly accurate in silico bioinformatics tool for predicting protein solubility from protein sequence is sought. In this study, we developed a deep learning sequence-based solubility predictor, DSResSol, that takes advantage of the integration of squeeze excitation residual networks with dilated convolutional neural networks. The model captures the frequently occurring amino acid k-mers and their local and global interactions, and highlights the importance of identifying long-range interaction information between amino acid k-mers to achieve higher performance in comparison to existing deep learning-based models. DSResSol uses protein sequence as input, outperforming all available sequence-based solubility predictors by at least 5 percent in accuracy when the performance is evaluated by two different independent test sets. Compared to existing predictors, DSResSol not only reduces prediction bias for insoluble proteins but also predicts soluble proteins within the test sets with an accuracy that is at least 13 percent higher. We derive the key amino acids, dipeptides, and tripeptides contributing to protein solubility, identifying glutamic acid and serine as critical amino acids for protein solubility prediction. Overall, DSResSol can be used for fast, reliable, and inexpensive prediction of a protein's solubility to guide experimental design.

Sequence-Based Explainable Hybrid Song Recommendation

Despite advances in deep learning methods for song recommendation, most existing methods do not take advantage of the sequential nature of song content. In addition, there is a lack of methods that can explain their predictions using the content of recommended songs and only a few approaches can handle the item cold start problem. In this work, we propose a hybrid deep learning model that uses collaborative filtering (CF) and deep learning sequence models on the Musical Instrument Digital Interface (MIDI) content of songs to provide accurate recommendations, while also being able to generate a relevant, personalized explanation for each recommended song. Compared to state-of-the-art methods, our validation experiments showed that in addition to generating explainable recommendations, our model stood out among the top performers in terms of recommendation accuracy and the ability to handle the item cold start problem. Moreover, validation shows that our personalized explanations capture properties that are in accordance with the user’s preferences.